{
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\begin{tabular}{l*{4}{c}}
\hline\hline
            &\multicolumn{1}{c}{(1)}         &\multicolumn{1}{c}{(2)}         &\multicolumn{1}{c}{(3)}         &\multicolumn{1}{c}{(4)}         \\
\hline
Used a booth to vote&    -0.00930         &     -0.0337\sym{**} &     -0.0138         &     -0.0162         \\
            &    (0.0150)         &    (0.0167)         &    (0.0163)         &    (0.0174)         \\
[1em]
Voted for PP&     -0.0258         &     -0.0366         &     -0.0371         &     -0.0391         \\
            &    (0.0264)         &    (0.0297)         &    (0.0270)         &    (0.0296)         \\
[1em]
Used booth x voted PP&       0.106\sym{**} &       0.118\sym{**} &       0.105\sym{**} &       0.107\sym{**} \\
            &    (0.0431)         &    (0.0480)         &    (0.0423)         &    (0.0460)         \\
\hline
Controls    &          No         &         Yes         &          No         &         Yes         \\
Region fixed effects&                     &                     &                     &                     \\
\hline\hline
\multicolumn{5}{l}{\footnotesize Standard errors in parentheses}\\
\multicolumn{5}{l}{\footnotesize Standard errors are robust}\\
\multicolumn{5}{l}{\footnotesize The outcome variable is a dummy for whether each respondent showed}\\
\multicolumn{5}{l}{\footnotesize signs of discomfort during the survey interview}\\
\multicolumn{5}{l}{\footnotesize Models 2 and 4 include controls for income, education, age, age squared, size of}\\
\multicolumn{5}{l}{\footnotesize respondent's municipality, and a dummy for respondents identifying as female}\\
\multicolumn{5}{l}{\footnotesize \sym{*} \(p<0.10\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\)}\\
\end{tabular}
}
